Artificial intelligence in pediatrics: a bibliometric analysis of global output, networks, and frontiers [2016–2025]

Background Although artificial intelligence (AI) is increasingly used to prevent, diagnose, and treat pediatric diseases, a quantitative panorama of the global research landscape is lacking. This study aimed to map the evolution of AI-focused pediatric literature, identify leading actors and themes, and detect emerging directions over the past decade. Methods Publications from 2016 to 2025 were retrieved from the Web of Science Core Collection (WoSCC) and analyzed with CiteSpace 6.4 R1 for volume trends, citation impact, collaboration networks (country/institution), keyword co-occurrence, burst detection, and co-citation clusters. Results Of 1,893 articles, annual output rose from 53 to 526 (R2=0.93). Retinopathy of prematurity (ROP), pediatric pneumonia, and prenatal congenital heart disease (CHD) formed the most cited knowledge clusters. The USA dominated productivity (659 papers), while France exhibited the highest betweenness centrality (0.19) despite ranking 11th in volume, reflecting historical European collaboration networks rather than active domain expansion. Keyword bursts since 2020 signal growing interest in virtual reality, growth analytics, and neonatal respiratory distress syndrome, with a notable shift toward explainable AI and outcome-focused research. Co-citation timelines reveal a shift from algorithm-driven studies to outcome-focused and explainable AI research. Conclusions AI applications in pediatrics are expanding rapidly, but collaboration remains institutionally fragmented. Connectivity, rather than mere productivity, underpins global influence; future efforts should adopt domain-specific strategies: prospective validation for mature imaging domains, data standardization for inconsistent areas, and ethical frameworks for emerging technologies, to accelerate safe clinical translation.